Evolving Genes to Balance a Pole

TL;DR

This paper explores using a Genetic Regulatory Network as an evolutionary approach to solve the pole balancing problem, demonstrating its ability to generalize across the domain.

Contribution

It introduces a modified Artificial Regulatory Network for reinforcement learning and analyzes its effectiveness and generalization capabilities.

Findings

Network successfully balances the pole

Representation generalizes well across problem variations

Different models of the network show varied performance

Abstract

We discuss how to use a Genetic Regulatory Network as an evolutionary representation to solve a typical GP reinforcement problem, the pole balancing. The network is a modified version of an Artificial Regulatory Network proposed a few years ago, and the task could be solved only by finding a proper way of connecting inputs and outputs to the network. We show that the representation is able to generalize well over the problem domain, and discuss the performance of different models of this kind.